Re: [Jamvm-general] Upcoming 2.4.2 Release - JamVM 2013-08-28 x86_64 update - improved jdk java lan
Brought to you by:
rlougher
From: Xerxes R. <xe...@za...> - 2013-09-17 17:50:54
|
2013-09-17 18:49, Andïï skrev: > On 16 September 2013 11:13, Xerxes Rånby <xe...@za...> wrote: >> 2013-09-13 16:43, Xerxes Rånby skrev: >>> 2013-09-12 16:58, Andïï skrev: >>>> On 12 September 2013 15:11, Xerxes Rånby <xe...@za...> wrote: >>>>> 2013-09-11 22:25, Andïï skrev: >>>>>> Some of you may have seen that Oracle have released u40 [0]. I'm >>>>>> currently testing 2.4.2, which will bring the 2.4 branch up to date >>>>>> with upstream on u40: >> ... >>>>>> I plan to release this coming Wednesday, the 18th of September, if all >>>>>> looks good by then. >>>>>> Xerxes is going to look at updating JamVM support on the branch in the meantime. >> ... >>>>> JamVM test logs, before jamvm update: ia32 & x86_64 >>>>> >>>>> Test procedure: >>>>> apt-get install mercurial >>>>> apt-get build-dep openjdk-7 >>>>> hg clone http://icedtea.classpath.org/hg/release/icedtea7-2.4 >>>>> cd icedtea7-2.4 >>>>> ./autogen.sh >>>>> ./configure --enable-jamvm --disable-system-lcms ; make ; make check >>>>> >>> ... >>>>> https://gist.github.com/xranby/a7fcbcbb08e0b80164fb - jamvm 2.4.2pre+r41272f3d7393 ia32 >>>>> https://gist.github.com/xranby/87bcbec586aca4ba8c70 - jamvm 2.4.2pre+r41272f3d7393 x86_64 >>> ... >>>> >>>> Thanks! I await the improved results from the new version :) >>> >>> >>> Updated logs for JamVM x86_86 using the latest 2013-08-28 revision >>> http://git.berlios.de/cgi-bin/cgit.cgi/jamvm/commit/?id=ac22c9948434e528ece451642b4ebde40953ee7e >>> >>> Logs: https://gist.github.com/xranby/d8f6c5f16e0f456a0f43 - jamvm 2.4.2pre+r41272f3d7393+ 2013-08-28 x86_64 >>> Patch: >>> http://git.berlios.de/cgi-bin/cgit.cgi/jamvm/commit/?id=ac22c9948434e528ece451642b4ebde40953ee7e >>> >> ... >>> >>> Improvement seen after the update: >>> >>> diff -Naur 7c8dceb90880616b7dd670f257961a1f5f371ec3 ac22c9948434e528ece451642b4ebde40953ee7e >>> --- 7c8dceb90880616b7dd670f257961a1f5f371ec3 2013-09-13 16:22:43.909428937 +0200 >>> +++ ac22c9948434e528ece451642b4ebde40953ee7e 2013-09-13 16:22:14.889285035 +0200 ... >>> -Test results: passed: 193; failed: 70; error: 3 >>> +Test results: passed: 196; failed: 67; error: 3 >>> >>> Three more hotspot tests passed after the update. >> ... >>> --------------- jtreg console summary for jdk --------------- >> ... >>> I have issues running the JDK tests on this x86_64 machine in general. >>> The JDK tests abort without completing using both Hotspot and JamVM. >> >> I have run the JDK jtreg jdk java lang tests manually in order to see the new JamVM JSR 292 improvements after the update. >> ./openjdk.build/j2sdk-image/jre/bin/java -jar test/jtreg.jar -v1 -s openjdk/jdk/test/java/lang | tee java-lang.log >> >> Summary of the diff in jdk java lang test results using jamvm on x86_64, before and after the update: >> --- 7c8dceb90880616b7dd670f257961a1f5f371ec3-icedtea-2.4-jamvm-java-lang.log 2013-09-16 11:56:01.539380859 +0200 >> +++ ac22c9948434e528ece451642b4ebde40953ee7e-icedtea-2.4-jamvm-2013-08-28-java-lang.log 2013-09-16 11:54:44.426998476 +020 ... >> -Test results: passed: 260; failed: 96; error: 22 >> +Test results: passed: 267; failed: 88; error: 23 >> >> Thus 7 more tests pass, no regressions seen in jdk java-lang after the update. >> The FAILED -> Error test of java/lang/invoke/CallSiteTest is IMHO OK for JamVM since this CallSiteTest tests for an Hotspot specific Indify optimization in order to pass. >> I recommend that we push this new 2013-08-28 update to HEAD and merge into icedtea-2.4 before the 2.4.2 release. >> >> Cheers >> Xerxes >> > > So this is newer than what's in HEAD? > > Ok, then please push to HEAD and I'll merge both changesets over. > Ok, I have pushed the new update to IcedTea 7 HEAD. Please merge the two change-sets into 2.4. http://icedtea.classpath.org/hg/icedtea7/rev/731375f0ba2e - JamVM: Fix issue with recent versions of JEdit; Updated to 2013-08-28 revision. 2013-09-17 Xerxes Rånby <xe...@za...> Robert Lougher <ro...@ja...> JamVM - JSR 335: fix frame.c - Add support for CallerSensitive - JSR 335: handle interfaces when resolving a method - sun.misc.Unsafe: additional methods get/putAddress - JSR 292: make internal functions static * NEWS: Updated. Removed note about JSR 355 for IcedTea 2. JSR 335 is only enabled for IcedTea 3 / OpenJDK 8 builds. * Makefile.am (JAMVM_VERSION): Updated JamVM to 2013-08-28 revision. (JAMVM_SHA256SUM): Updated. One important part of this update is the fixed recent JEdit version support found and fixed by Robert. http://git.berlios.de/cgi-bin/cgit.cgi/jamvm/commit/?id=74ebffd419716c9d2779fc1b9cec14cc51af4820 - sun.misc.Unsafe: additional methods get/putAddress I Removed the erroneous NEWS about JSR 335, it was never enabled for IcedTea 2. It was erroneous added to the icedtea 2 NEWS in the 2013-07-14 revision update. This JamVM commit part of the 2013-07-14 revision update http://git.berlios.de/cgi-bin/cgit.cgi/jamvm/commit/?id=1a58072f8339270f9372c273d82a00790e04375d - JSR 292: enable for OpenJDK 7/IcedTea 2 Clearly show JSR 335 is only enabled for IcedTea 3 / OpenJDK 8 builds. http://icedtea.classpath.org/hg/icedtea7/rev/b5fa79fec364 - JamVM: JSR 335: Lambda Expressions; JSR 292: enable for OpenJDK 7/IcedTea 2; Updated to 2013-07-14 revision. 2013-08-14 Xerxes Rånby <xe...@za...> Robert Lougher <ro...@ja...> JamVM - JSR 292: enable for OpenJDK 7/IcedTea 2 - JSR 335: invokeinterface check in method resolution - JSR 335: java.lang.invoke.MagicLambdaImpl - JSR 292: invokedynamic is 5 bytes - non-direct interp - JSR 292: invokedynamic is 5 bytes in length - Minor formatting change - Minor fixes: use perror for mmap failure - FreeClassData: fix comment - FreeClassData: adjust method count for Miranda methods - JSR 335: remove GC hack - JSR 335: implement "bridge" methods for Mirandas - JSR 335: handle multiple defaults and conflicts * NEWS: Updated. * Makefile.am (JAMVM_VERSION): Updated JamVM to 2013-07-14 revision. (JAMVM_SHA256SUM): Updated. Cheers Xerxes |